US 7272817 B1 Abstract Method and apparatus for modeling a business process to facilitate evaluation of driving metrics. A decision tool can be provided to help manage a business process. Model files are created based on data pertaining to a plurality interrelated metrics. Each of the plurality of model files is optimized. The results of the optimization for the various metrics can be combined into a summary, which describes a hierarchy of selected driving metrics in such a way as to facilitate an understanding how to fine-tune various metrics to meet goals. An embodiment of the invention can take the form of a stand-alone computing system running a spreadsheet program, a stand-alone computer system running a dedicated application, or a computer system interconnected with a data warehouse to acquire current data regarding the interrelated metrics.
Claims(27) 1. A method of modeling a business process to facilitate an evaluation of driving metrics for a selected goal metric, the method comprising:
gathering data to define a plurality of interrelated metrics associated with the business process;
creating a plurality of models based on the data, each of the plurality of models corresponding to one of the plurality of interrelated metrics;
performing a non-linear optimization for each of the plurality of models to adjust the primary coefficients and exponents to minimize the difference between an estimated value and a goal value for the selected goal metric; and
running a back-substitution routine to define the selected goal metric in terms of the substantially smallest to the substantially largest of the selected driving metrics to combine the results of the non-linear optimization for each of the plurality of models to produce a summary which describes a hierarchy of selected driving metrics for the selected goal metric.
2. The method of
3. The method of
setting a correlation coefficient starting value to limit the number of selected driving metrics to a preselected maximum number; and
omitting from the creating of the plurality of models the interrelated metrics having a correlation coefficient that is less than a preselected starting value.
4. The method of
5. The method of
setting a correlation coefficient starting value to limit the number of selected driving metrics to a preselected maximum number; and
omitting from the creating of the plurality of models the interrelated metrics having a correlation coefficient that is less than a preselected starting value.
6. The method of
7. The method of
8. A computer readable storage medium comprising a computer program for modeling a business process to facilitate an evaluation of driving metrics for a selected goal metric, the computer program further comprising:
instructions for gathering data to define a plurality of interrelated metrics associated with the business process;
instructions for creating a plurality of models based on the data, each of the plurality of models corresponding to one of the plurality of interrelated metrics;
instructions for performing a non-linear optimization for each of the plurality of models to adjust the primary coefficients and exponents to minimize the difference between an estimated value and a goal value for the selected goal metric; and
instructions for running a back-substitution routine to define the selected goal metric in terms of the substantially smallest to the substantially largest of the selected driving metrics to combine the results of the non-linear optimization for each of the plurality of models to produce a summary which describes a hierarchy of selected driving metrics for the selected goal metric.
9. The computer readable storage medium of
10. The computer readable storage medium of
instructions for setting a correlation coefficient starting value to limit the number of selected driving metrics to a preselected maximum number; and
instructions for omitting from the creating of the plurality of models the interrelated metrics having a correlation coefficient that is less than a preselected starting value.
11. The computer readable storage medium of
instructions for setting a correlation coefficient starting value to limit the number of selected driving metrics to a preselected maximum number; and
instructions for omitting from the creating of the plurality of models the interrelated metrics having a correlation coefficient that is less than a preselected starting value.
12. The computer readable storage medium of
13. The computer readable storage medium of
14. The computer readable storage medium of
15. Apparatus including a processor to facilitate an evaluation of driving metrics for a selected goal metric, the apparatus comprising:
means for gathering data to define a plurality of interrelated metrics associated with the business process;
means for creating a plurality of models based on the data, each of the plurality of models corresponding to one of the plurality of interrelated metrics;
means for performing a non-linear optimization for each of the plurality of models to adjust the primary coefficients and exponents to minimize the difference between an estimated value and a goal value for the selected goal metric; and
means for running a back-substitution routine to define the selected goal metric in terms of the substantially smallest to the substantially largest of the selected driving metrics to combine the results of the non-linear optimization for each of the plurality of models to produce a summary which describes a hierarchy of selected driving metrics for the selected goal metric.
16. The apparatus of
17. The apparatus of
18. The apparatus of
19. The apparatus of
20. A system for modeling a business process to facilitate an evaluation of driving metrics for a selected goal metric, the system comprising:
a user input screen to receive as input, data to define a plurality of interrelated metrics associated with the business process, and parameters for controlling the manner in which some of the plurality of interrelated metrics are reflected in a summary describing a hierarchy of selected driving metrics for the selected goal metric; and
a processing platform including a computer readable medium whose contents cause the processing platform to create a plurality of models corresponding to the plurality of interrelated metrics, to perform a non-linear optimization for each of the plurality of models to adjust primary coefficients and exponents to minimize the difference between an estimated value and a goal value for the selected goal metric, and to run a back-substitution routine to define the selected goal metric in terms of the substantially smallest to the substantially largest of the selected driving metrics to produce a summary which describes a hierarchy of selected driving metrics for the selected goal metric.
21. The system of
22. The system of
23. The system of
24. The system of
25. The system of
26. The system of
27. The system of
Description A portion of the present disclosure is contained in a compact disc, computer program listing appendix. The compact disc contains an MS-DOS text file named B-Engine.txt created on Mar. 25, 2004, of approximately 83 kilobytes. The contents of this file are incorporated herein by reference. Any references to “the appendix” or the like in this specification refer to the file contained on the compact disc. The contents of this file are subject to copyright protection. The copyright owner has no objection to the reproduction by anyone of the patent document or the appendix as it appears in the Patent and Trademark Office patent files or records, but does not waive any other copyright rights by virtue of this patent application. Understanding how to execute a business process within a company or enterprise in order to maximize revenue, profit, or other metrics, is of enormous importance and has a significant impact on the company's success in the marketplace. Ideally therefore, business processes should be monitored, modeled, and optimized in much the same ways as scientific or manufacturing processes. In particular, it is important to know how various activities within the business process drive various metrics. Thus, it can be understood how to adjust various subprocesses and their drivers in order to achieve appropriate performance relative to one or more overriding goals. Despite the above described need, it is often difficult to model and describe the activities and metrics that are part of a business process in a way that accurately reflects their relationships to each other and their effects on one another. Very often, the various activities included in a business process are developed independently and no one person has the comprehensive knowledge necessary to accurately determine how one metric affects another. Often, data which attempts to characterize these relationships is centered around separate systems and lacks currency. In other cases, more comprehensive modeling and evaluation can be accomplished, but only with extensive, laborious, calculations performed over an extended period of time. The present invention, as illustrated by example embodiments, can provide a decision tool that helps manage the complexities of a business process in order to maximize or improve a goal metric. In many cases, the goal metric might be profitability, customer satisfaction, income, or other metrics, which tend to be of overriding importance to a business. However, the techniques disclosed can be used to maximize or optimize any metric. A model or models, which can be created through use of embodiments of the invention, can be used to find optimal values for selected driving metrics for a goal metric. Such driving metrics can include, for example, such things as staffing levels, processing times, shrinkage, and other performance metrics in regard to providing goods or services. The modeling techniques of the invention can be used to provide more definitive, accurate, and timely information on how to optimize business performance in any organization or enterprise. In example embodiments of the invention, a business process can be modeled to facilitate an evaluation of driving metrics which drive or can drive a selected goal metric. The modeling begins by gathering data to define a plurality of interrelated metrics associated with the business process. The data can include parameters for controlling the manner in which the metrics are modeled and reflected in the analysis. The data can also include a listing of all the interrelated metrics and their known or average values. In some embodiments, the later can be gathered via connections to data stores or a data warehouse. A plurality of model files based on the data is created, at least in part through regression analysis and nonlinear optimization. Each of the model files corresponds to one of the interrelated metrics. In example embodiments, each model is optimized to set coefficients and exponents to optimal values. The optimal values can be determined by minimizing the difference between an estimated value and a goal value for the selected goal metric. Finally, in example embodiments, the results of the optimization for the various metrics are combined into a summary which describes a hierarchy of selected driving metrics. Parameters that can be input prior to performing an analysis can include a maximum number of driving metrics for each level. Other parameters can include a maximum value for the probability that a coefficient is zero, which can also be used to limit the number of metrics which are included in the summary. The controllability of a metric can also be indicated. Thus, a metric which is not easily controllable by the business can be included in a model using a constant value to represent the metric. In example embodiments, the production of the summary report can include running a back substitution routine to further define the selected goal metric in terms of the smallest to the largest of the selected driving metrics. The back substitution routine can be followed by another nonlinear optimization to further define the selected goal metric in terms of the selected driving metrics. Additionally, time lag effects of various metrics can be taken into account in the summary based on time lag information input prior to an analysis being run. In some embodiments, the invention is implemented via either a stand-alone computing platform or a computing platform interconnected with other platforms or data stores by a network, such as a corporate intranet, a local area network, or the Internet. A computer program product or computer program products contain computer programs with various instructions to cause the hardware to carry out, at least in part, the methods and processes of the invention. Data stores or a data warehouse can be connected to a computing platform that is performing the analysis. Dedicated software can be provided to implement the invention, or alternatively, a spreadsheet program with appropriate macros can be used to implement embodiments of the invention. In either case a user input screen is operable to receive appropriate input for controlling the manner in which the interrelated metrics are reflected in the summary, and a processing platform creates the models, performs nonlinear optimization, and combines the results in the appropriate manner. The present invention can most readily be understood by considering the detailed embodiments presented herein. Some of the embodiments are presented in the context of an enterprise using software to facilitate modeling and analysis in order to determine how to best optimize specific business metrics. However, these embodiments are examples only. It cannot be overemphasized that the invention has applicability to any type or size of organization and can be used to optimize any type of metrics. The present invention can be embodied in computer software or a computer program product. An embodiment may include a spreadsheet program and appropriate macro programs, algorithms, or plug-ins. An embodiment may also consist of a custom-authored software application for any of various computing platforms. One specific example discussed herein involves the use of a Windows™ personal computing platform running Microsoft Excel™ spreadsheet software, with appropriate Visual Basic™ macros. It cannot be overemphasized that this embodiment is an example only. The source code for example Visual Basic macros, which enable the invention to be implemented in such an example embodiment is included in the appendix. The source code example will be readily understood by those of ordinary skill in the art. It will also be readily understood that the inventive concepts described herein can be adapted to any type of hardware and software platform using any operating system including those based on Unix™ and Linux. In any such embodiments, the instruction execution or computing platform in combination with computer program code instructions form the means to carry out the processes of the invention. Some of the terminology used in this description should be understood from the beginning. The term “metric” and similar terms refer to measurable characteristics of a business process that can be expressed in numerical or mathematical terms. All of the metrics related to a particular business process may be referred to herein as “interrelated metrics” or the like. A “selected goal metric” or simply a “goal metric” is a metric from among the various interrelated metrics for a business process that a user of an embodiment of the invention is seeking to optimize. The models created with an embodiment of the invention can be used to determine which other metrics drive a selected goal metric. Such other metrics are referred to as “driving metrics” or the like. It should be noted that the selected goal metric is one of the interrelated metrics for a given business process model. As such, it is possible to run the same analysis repeatedly or concurrently with the same set of interrelated metrics and the same set of models but selecting a different metric as the goal metric for each analysis. Terms such as “selected driving metric” and the like refer to the interrelated metrics that are reflected in a summary of an analysis. In some cases some interrelated metrics are eliminated during analysis. Thus, the selected driving metrics can be all of the interrelated goal metrics. However, as will become clear, example embodiments of the invention include mechanisms to eliminate certain interrelated metrics because their effect is minimal, or their presence would complicate an analysis. The term “coefficient” when used by itself is meant in the generic mathematical sense, that is, when equations are used to model a process, the coefficients are the numbers by which various metric values are multiplied in an equation. A “correlation coefficient” is also used in certain embodiments of the invention, and this term is meant in its statistical sense as distinct from the generic term “coefficient.” An understanding of the invention may be facilitated by an understanding of the way a business process can be modeled as a relationship where one metric is driven by other metrics, which are in turn driven by other metrics, etc. Thus, a selected goal metric is driven by a certain set of metrics. Each metric from that certain set of metrics is also driven in turn by other metrics. Such a relationship can be illustrated by a set of equations like that shown below. The selected goal metric is designated “Y” and might sometimes be referred to herein as the “big Y” metric. The big Y metric is a function of a plurality of driving metrics, designated in the first equation as x's. The second and third equations express how each of the x's that drive the big Y are in turn driven by other x's, and the equations can continue to cover the other x's in the first equation. Then, additional equations show that the x's from the second level of equations are in turn driven by still other x's, and so on.
The relationships expressed above can continue through as many orders as one would like, at least for complex business processes. As will be seen below, in example embodiments of the invention, mechanisms are in place to limit the number of selected driving metrics which are used to model a particular business process for analysis. However, the extent to which these metrics are limited can be determined by an individual user of the invention. Embodiments of the invention can be used to model an infinite variety of business processes. For purposes of an illustrative example presented herein, it is assumed that the operating process of a customer call center is being analyzed. The analysis can help determine how to drive a particular goal metric by changing driving metrics such as the telephone call answer rate, how much authority to resolve a problem do the personnel manning the telephones have, amount of overtime worked, etc. The invention can be applied to any other business process. For example, an analysis could be done to determine how to maximize retail sales based on driving metrics such as the number of hours a store is open each day, the number of sales people available at any given time, etc. Similarly, a manufacturing process could be analyzed in terms of number of workers per shift, quantity of parts available, etc. At block Additional parameters which can be input according to certain embodiments of the invention are a goal value for the selected goal metric, a “p-limit” for use in calculations, a starting value for a correlation coefficient, and a maximum number of selected driving metrics to be included in the analysis. The “p-value” is a probability associated with the risk of taking action because things appear different when they are in fact the same. This value establishes the threshold for what are significant factors as opposed to what are trivial factors in driving the selected goal metric. In statistical terms, the p-value for a given metric is the probability that its coefficient is zero, that it in fact has no effect on the goal metric or the metric above it. The “p-limit” is the p-value that must be reached before a specific metric is ignored in the analysis. A smaller p-limit increases the risk that an important metric will be ignored; a larger p-limit decreases that risk, but increases the number of selected driving metrics that will be included, and hence, the complexity of the models in the analysis. The starting correlation coefficient or “r-start” indicates the amount of correlation (“r”) that must be achieved for a given metric before it is included in the analysis. The correlation between variables establishes a threshold for sets of equations. Correlation in this context is used in the statistical sense as is understood in the statistical arts. Requiring a large correlation coefficient for a driving metric to be included in an analysis increases risk but also decreases the number of driving metrics included in the analysis and in an eventual summary report. The maximum number of driving metrics allows a user to limit the size of the analysis and the summary report to make it more convenient to work with. As will be seen below, the starting correlation coefficient is used in combination with the maximum number of drivers to allow a system implementing the invention to create a model and a summary which is of the desired size and complexity for a given user. Returning to From the data input, an instruction execution system fits a regression line and creates an equation hierarchy that describes the relationship between all of the metric variables for the current model. At block At block Before delving into the details of the optimizations performed by an embodiment of the invention, it may be helpful for the reader to see how the various input data is organized in a user input screen. Optional parameters are input in fields The main part of user input screen The setup area, In drop-down box At block At block Column Section Processing platform In any case, a computer program which implements all or parts of the invention through the use of systems like those illustrated in The appendix to this application includes Visual Basic source code for a collection of macros which work with the well-known Microsoft Excel spreadsheet program. The source code can be used by one of skill in the art to cause a computer system implementing Microsoft Excel to carry out the methods and processes of an example embodiment of the invention. The text file contains source code for four macros all listed one after the other, however; each one is delineated by appropriate text headings. The macros consist of control panel code which implements a user input screen, a main program which contains common routines which are used throughout the various stages of executing an embodiment of the invention, and two macros which make use of the standard solver tools which are included within at least the “XP” versions of Microsoft Excel. One of the macros includes solver fitting code, and the other includes solver target code. The macros can be used to produce a Microsoft Excel spreadsheet and appropriate Excel add-ins. It cannot be over-emphasized that the exact implementation implicated by the appendix to this application is but an example. One of ordinary skill in the art can easily adapt the principles learned from studying the source code to other systems, dedicated applications, and even other computing platforms which do not make use of Microsoft Excel or any other spreadsheet program. Specific embodiments of an invention are described herein. One of ordinary skill in the computing and statistical arts will recognize that the invention can be applied in other environments and in other ways. It should also be understood that not all of the elements and features in the drawings, or even in any one drawing are necessary to implement the invention as contemplated by any of the appended claims. Likewise, an implementation of the invention can include features and elements or steps in addition to those described and claimed herein. Also, the steps in the appended claims are not necessarily conducted in the order recited, and in fact, can be conducted in parallel in some cases. Thus, the following claims are not intended to limit the scope of the invention to the specific embodiments described herein. Patent Citations
Referenced by
Classifications
Legal Events
Rotate |